Skip to main content

Full text of "The MAGSAT vector magnetometer: A precision fluxgate magnetometer for the measurement of the geomagnetic field"

See other formats


NASA 

TM 

79656 

c.l 



Technica! yemorandum 7965S 

(NASA-TM-796 56) THE MfiGSP.I. VECTOB 
HAGHETOKETEB; R FHECISIOK FLOSGATE 
MAGHITOMETEB FOR THE HEaSOl^EHlHT OF THE 
GEOHAGHETIC EIELD {N&S&) 33 p HC A03/MF R01 

CSCl 08N G3/ft6 

T^.© MAQSAT Vector 



LOAN COPY: R 
AFWL TECHNif a 
KIPJLAMO A^Etn 



N79- 16520 



anclas 

17936 



XT ; 



m 

o 



> 



S 



ifietometer 



i\i 



«> I 



^HsOs 









&E 



e 



Vissf M B 



sB H^!*' 



l<*s« S 



6 SB'" ft"™ (3 es»a 






*^M»5 



l^ge 



4;;i*"'% 



>V 










October 1 971 



National Aeronautics and 
Space Administration 

Goddard Space Flight Center 
Greenbelt, Maryland 20771 




:y 



V, .■-:. 



TECH UIHARY KAFB, m 




DlSDMb? 



THE MAGSAT VECTOR MAGNETOMETER - A PRECISION FLUXGATE 
MAGNETOMETER FOR THE MEASUREMENT OF THE GEOMAGNETIC FIELD 



by 



M.H. Acuna, C.S. Scearce, J.B. Seek, J. Scheifele 

Planetary Magnetospherer; Branch 

Laboratory for Extraterrestrial Physics 

NASA/Goddard Space Flight Center 

Greenbelt, Maryland 20771 



ABSTRACT 

A description of the precision triaxial fluxgate magnetometer to be 
flown aboard the MAGSAT spacecraft is presented. The insurument covers 
the range of + 64,000 nT with a resolution of + 0.5 nT, an intrinsic 
accuracy of + 0.001% of full scale and an angular alignment stability 
of the order of 2 seconds of arc. It was developed at NASA's Goddard 
Space Flight Center and represents the state-of-the-art in precision 
vector magnetometers developed for spaceflight use. 
INTRODUCTION 

The general objectives of the MAGSAT mission are the acquisition 
of accurate magnetic field data from a low polar orbiting spacecraft to 
obtain an accurate and up-to-date quantitative description of the geo- 
magnetic field as well as the compilation of global scalar and vector 
crustal magnetic anomaly maps. The instrumentation aboard the MAGSAT 
spacecraft consists of an alkali vapor scalar magnetometer, a precision 
vector magnetometer and an attitude determination system composed of 
two star cameras, a precision sun sensor and a boom attitude transfer 
system. A more detailed description of the mission and its objectives has 
been given by Langel"^ ' . 

The accuracy goals for the mission require a vector magnetometer 
capable of measuring the geomagnetic field with a maximum error of + 1 nT 
in magnitude and 5 seconds of arc in direction. The development of such 
an instrument within the constraints imposed by the spacecraft and its 
subsystems represents a major technological achievement which would not 
have been possible without parallel developments in the areas of ultrapreclsion 
linear integrated circuits and passive components. The design presented hero 



represents a compromise among many conflicting requirements and limitations 
imposed by reliability considerations, available resources and state-of-the- 
art components. Other implementation schemes are undoubtedly possible with 
equivalent performance but their incompatibility with the spacecraft and 
other instruments precluded their utilization. 
INSTR U MENTATION 

A block diagram of the Vector Magnetometer is shown in Figure 1. 
The instrument consists of a basic triaxial fluxgate magnetometer with a 
dynamic range of + 2000 nT, a 12-bit analog-to-digital converter and 
three 7-bit digital-to-analog converters which are used as field offset 
generators for each orthogonal axis to increase the measurement dynamic 
range to + 64,000 nT in steps of + 1000 nT. The bias steps are automatically 
added or subtracted depending on the magnitude of the external field, to 
maintain the fluxgate magnetometer within its operating range, as shown in 
Figure 2. 

This approach yields +0.5 nT resolution and is equivalent to utilizing 
a magnetometer with a + 64,000 nT dynamic range in conjunction with a 
17-blt analog-to-digital converter. The reasons for choosing this 
"differential measurement" approach are basically considerations of 
dynamic response of a 17-bit A/D converter, magnetometer feedback loop 
stability and slew rate, as well as zero level offset sta"-ility versus 
a.c. gain. Some of these tradeoffs are presented below where the 'n.ip.r.e'- oniet>3r 
electronics and transfer function arc discussed. The external magnet ic 
field is not expected to change appreciably bet^^'een samples and the 
differential approach with a basic "window width" of + lOOO nT is '.nore 
than adequate. 



The magnetic field is sampled 16 times per second along each of the 
three orthogonal directions and the field offset generator is updated 
at the same rate. Thus, if we assume that a 64,000 nT field is suddenly 
applied to the sensor, it takes approximately 4 seconds (64 x .0625) for 
the field offset generator to step to the proper level and thus obtain 
a valid measurement. However, this is a worst case situation since we are 
dealing with a non-spinning spacecraft and the external field does not 
vary significantly between samples. 

The digital data corresponding to the 'fine' (12 bits) and 'coarse' 
(7-bits) readouts are fed directly to the spacecraft through a serial 
interface. No on-board processing of data is utilized. 

As can be seen from Figure 1, the magnetometer electronics, A/D 
converter, field offset generators, digital interface and power converter 
are implemented in a redundant configuration which can be selected by 
ground command in case of failures. This block redundancy approach 
provides a fundamentril measure of reliability and e]iminates from con- 
sideration single poin;- failures in the electronics. 

The digital interface logic, 12-bit A/D converter and power converter 
are irnrlemented v/ith cooventi»)nal designs using CMOS mi croclrcuits and 
switching regulators r.v.spec tively , and will not be covered here. 

The fluxgate sensor assembly and associated electronics and the 
field offset generator design are presented below. The last section 
covers the calibration and alignment procedures utilized to verify 
instrument performance. 



\ 



TRIAXIAL FLUXGATE SENSOR 

The fluxgate sensors are constructed utilizing the ring core geometry 
which has been shown to exhibit superior performance characteristics in 
terms of drive power requirements, long term zero stability and noise level 
when compared to other types of fluxgate sensors. The mag-.ietic material 
used in these sensors is an advanced molybdanum permalloy especially 
developed for the GSFC-Voyager Magnetic Field Experiment in cooperation 
with the Naval Surface Weapons Center, White Oak, Maryland. It exhibits 
extremely low noise and high stability characteristics typical of 6-81 
Mo.Pe. alloys developed for fluxgate magnetometer applications by the 
NSWC group. The use of these alloys and the ring core sensor geometry 
allows the realization of compact, ultrastable fluxgate sensors with 
unmatched performance. 

A schematic drawing showing the construction of each sensor is sho'Am 
in Figure 3 while the triaxial sensor assembly is shown in Figure 4. 
Crucial elements in the design of the MAGSAT sensor are the feedback 
coil v;hich nulls the ambient field and the sensor core itself since they 
can be shown to be the most significant sources of error, both from 
alignment stability point of view as well as the scale factor stability 
versus temperature. 

The approach utilized here is to use individual axis nulling, "hat is, 
each axis is nulled independently in only one direction. This is the 
most common approach but suffers from extreme sensitivity to cross-fields 
since the ambient field is nulled only along the sensor axis. Thus any 
distortion or motion of the sensor core within the feedback coil represents 
an effective alignment change due to the presence of the transver.'je field. 



An approach that minimizes this sensitivity is to use a relatively 
large triaxlal Helroholtz coil system around the triaxial sensor such that 
the external field is nulled over the entire sensor volume rather than just 
along each axis individually. However, the large dimensions of the coil 
system introduce additional complications in terms of alignment stability, 
thermal control and particularly in the case of the MAGSAT spacecraft, 
the generation of relatively large fields at the location of the scalar 
magnetometer due to the feedback current. Hence, the individual axis 
nulling approach was selected for implementation. 

A rigid attachment of the sensor core to the feedback coil usually 
results in differential expansion stresses being transmitted to the 
sensor core which can produce significant alignment shifts, zero level 
changes as well as dramatic increases in sensor noise levels. These 
problems can be solved by carefully matching the thermal expansion 
coefficients of the elements involved (core, windings, support structure) 
to minimize differential stresses and changes in the relative geometry 
between the core and the feedback coil. 

In Figure 5 we show the relative variation of the feedback field 
versus relative dimensional changes, both isotropic and anisotropic, 
introduced by temperature variations for a coil having a geometry similar 
to that utilized in the MAGSAT sensors, and a variety of structural 
materials. The computed field is that at the geometrical center of the 
coil but it can be easily shoim that the field intensity at any point 
within the coil follows a similar variation as a function of temperature. 
From the figure and formulas it can be seen that the dominating factor 
is the variation of coil length with temperature. Since the sensor 
responds to a weighted average of the magnetic field over an undetermined 
volume in space, the proceeding calculation constitutes a zeroth-order 
approximation to the actual behavior of the sensor. Nevertheless, 
the general temperature dependence shown in Figure 5 was verified 
for prototype instruments utilizing a 50,000 nT field "trapped" 
in a 15 cm superconducting shield and varying the sensor temperature 
over -40°C to +60°C. 



The fundamental strategy followed for MAGSAT was to match the 
expansion coefficient of the support structure and windings to that of 
the sensor core itself (~ 11x10 / C) . Of the structural materials shown 
in Figure 5, MACOR* was selected due to its expansion coefficient (10x10 ) 
zero water absorption (and thus dimensional stability versus relative 
humidity changes) and ease of machinability. The feedback coil was 
wound with polyur ethane-nylon insulaced platinum wire %-;ith a linear 
thermal expansion coefficient of 9.9 x 10 / C. 

In terms of alignment stability the small dimensions of the sensor 
require structural deformations to be smaller than 51 microns under 
all environmental conditions to obtain the desired alignment stability 
of 5 arc seconds. This extreme requirement necessitated the use of 

extremely stable, hysteresis free materials for the feedback coil and 

* 
support structure, such as MACOR , while the mounting surfaces were 

polished flat to a surface roughness of < 10 u prior to assembly. By 

closely matching the expansion coefficients of the sensor and feedback 

coil it was possible to rigidly bond these two components together and 

thus minimize relative displacements. The only remaining source of error 

is motion of the actual magnetic material within the core bobbin. This 

constituted a significant problem in spite of the close tolerances used 

during sensor core manufacture. The solution found was to increase the 

amount of magnesium oxide normally used to insulate the tape material , 

such as to fill the entire support bobbin and completely enclose the tape. 

This solution proved very effective in vibration and environmental "^ 

tests performed on a prototype sensor and no alignment shifts were 

detected. 

* Corning Glass Works trademark 



Since the overall expansion coefficient of the feedback coil is non- 
zero the magnitude of the feedback field, that is, the scale factor, 
will be a function of temperature as shown in Figure 5. In the magneto- 
meter electronics description given below, we show how this temperature 
dependence can be compensated for over a v^ide temperature range by means 
of a modified current source designed to produce a constant feedback 
field as a function of temperature. 

The external magnetic field generated by this sensor arrangement 
is less than + 1 nT at 46 cm from the sensor when measuring fields of 
+ 64,000 nT. This low field value minimizes the interference with the 
alkali vapor scalar magnetometer which is located only 43 cm away from 
the sensor assembly. 

As shown in Figure 4, two MACOR* optical cubes are bonded to the 
sensor mount to obtain accurate angular position determination during 
alignment tests described in the last section of this paper. The 
sensor assembly is mounted on a temperature stabilized baseplate which 
also supports two large mirrors associated with the Attitude Transfer 
System (ATS). Temperature stabilization of this baseplate is required 
to minimize optical alignment shifts between the sensor optical cubes 
and the ATS mirrors due to the different thermal expansion coefficients 
involved . 
MAGNETOMETHR ELECTRONICS 

The electronics design of the M/\GSAT Vector magnetometer incorporates 
a number of developments derived from the Voyaper Magnetic Field Experi- 
ment (Acunn,' '; Bchannon et al., )whLch aro directly applicable to 
the present instrument. Figure 6 shows a block diagram of the basic 



fluxgate magnetometer while a detailed schematic of the drive and signal 
processing electronics for each axis is shovm in Figure 7. A 250 kHz 
synchronization signal is derived from the spacecraft clock and divided 
down to 12.5 KHz to provide the excitation signal for the fluxgate sensors. 
A high efficiency 'capacitive discharge' circuit is utilized to drive 
the cores to over 100 times their coercive force with low average power 
requirements (Acuaa' ■■), The peak-to-peak sensor drive current is 
approximately 600 .nA while the average drive power requirement is only 70 mW. 
This arrangement essentially eliminates the problem of 'perming' of flux- 
gate sensors for externally applied fields of up to 20 gauss. The simple 
voltage regulator used in the driver stage ensures the stability of the 
drive signal amplitude over all environmental conditions. 

The signal processing electronics consist of an A.C. preamplifier 
tuned to the second harmonic of the drive signal, a synchronous 
demodulator, an operational integrator and a transconductance feedback 
summing amplifier which takes the place of the feedback resistor in 
conventional designs. A low pass, 6 db/octave filter is utilized at 
the output to limit the signal bandwidth to 8 Hz since as mentioned 
previously, the sampling rate is 16 samples/sec. 

The input circuit to the A.C. preamplifier incorporates a tuned 
secondary transformer while the primary side presents a low Impsdance 
to the fluxgate sensor. The inductance of the sense/ feedback sensor 
winding is series tuned with the«*input capacitor and reactive part of 
the input Impedance to obtain considerable amplification of the second 

harmonic signal in the sensor itself through negative resistance 

[2] 
parametric amplification (Acuna ) • The amount of parametric 



amplification obtained in the MAGSAT sensors is considerably smaller 
than in other similar fluxgate sensors due to the high re^iritance of 
the feedback winding which is constructed using insulated platinum 
wire. Nevertheless, the overall sensitivity achieved of 100 pV/nT 
at the secondary of the input transformer (compared to 10 mV/nT for 
copper wound coils) is more than adequate to obviate the need for 
an ultra-high gain A.C. preamplifier and its associated problems. 
A simple two-stage, hybrid preamplifier with a low noise FET at the 
input provides a gain of approximately 75. A low Q tuned circuit 
is Included at the output to provide additional rejection to unwanted 
odd frequency components which could affect the performance of the 
non- ideal syn':hronous detector. 

The low input impedance of the preamplifier essentially eliminates 
the problem of sensor cable length r.ffecting the tuning of the input 
stage and allows a significant Increase in the (2nd harmonic/fundamental) 
ratio applied to the first active device. All these features combine to 
produce an instrument remarkably free of 'electronic perm' and extremely 
temperature stable. 

The synchronous detector is a conventional design utilizing a 
quad CMOS transmission gate driven by a second harmonic reference 
signal derived from the frequency divider chain in the drive electronics. 
The operational Integrator provides the bulk of the gain in the loop 
and its time constant is selected to guarantee loop stability and the 
desired response bandwidth. The transfer function of the instrument 
is discussed in more detail below. 



10 



The output of the integrator is applied to the trasconductance 
amplifier and the output low pass filter. The amplifier is configured 
as a modified voltage controlled current source x,;hich sums in the 
proper ratio (32:1) the signals from the integrator and the field offset 
generator. This is the most crucial element in the design since it 
determines the accuracy and stability of the magnetometer scale factor. 
In the previous section we discussed the problem of scale factor 
dependence upon temperature due to thermal expansion of the feedback coil. 
In order to compensate for this effect we have taken advantage of the 
significant variation of feedback coil resistance with sensor temperature 
to maintain the feedback field constant. 

Since the variation of this field is, to a first approximation, 
mainly dependent upon coil length variations (see Figure 5) , we can write 



K I 

^ (1) 



"f id + cf T) 

M 

where I^ is the feedback coil current, K a proportionality constant, 
Z the coil length, a the thermal expansion coefficient of the support 
bobbin and coil wire (matched), and T the ambient temperature. In 
addition the feedback coil resistance varies with temperature as 

R = R (1 + a T) (2) 

CO p 

where R is the reference tempe -ature resistance value and a the thermal 
o p 

resistivity coefficient for the coll wire. 

Considering now the simplified schematic for the feedback amplifier 
shown in Figure 8, the current through the coil is given by 



11 



-(E + E ) 

f R,+R„-R ^ ' 

R, + R (1+a T) [ — — ] 



1 o p 



(R/^ + R2 



Introducing this equation in (1), differentiating with respect to T and 

ignoring second order terms of Ot'ttwa ), we find- that if we choose 

* n p 

B(R-R ) - (R/2) 
R. = -^-fg (4) 



where 



R a 

^ (1 + ^) (5) 



then the feedback field, as defined by Equation (1) will be independent 
of temperature. Basically, the circuit incorporates a slight amount of 
positive feedback to compensate for the decrease of Bp due to coil 
expansion. It is then possible to adjust R^, such as to make the scale 
factor dependence with temperature as small as possible. 

This compensation approach has been used with excellent results in 
several fluxgate variometers built at GSFC for ground based studies of 
the geomagnetic field. 

The operational amplifier used in this circuit is an ultrastable 
(0.1 uV/°C Wos) monolithic unit vjith excellent long term stability 
characteristics and very low noise. Other amplifier types were considered 
such as monolithic choppers, but were not used due to either noise 
performance, reliability or long term stability considerations. 



12 



t'^Jff.WM.^.tWM. I ^-l^^Wt^ 



p-ysy }"<■"> ' .■" -?-^8»»-y*i)ii ' 



The resistors associated with this circuit are ultraprecision, 
hermetically sealed units with an absolute temperature coefficient of 
< 0,5 ppm/ C. Tracking characteristics approach 0.2 ppm/ C, while 
long term stability after initial conditioning is of the order of 
20 pom/yr. 

Since the maximiim allowable error voltage at the amplifier input 
is only 125 yV (1 nT) , considerable attention was given to thermally 
generated e.ra.f.'s which can be produced across dissimilar metal 
junctions. The circuit layout is such that thermal gradients across 
critical paths are minimized by close proximity of conductors. Low 
thermal solder was used during assembly to further reduce the magnitude 
of the e.m.f.'s generated across solder joints. 

The detailed circuits shown in Ficjre 7 also incorporate means for 
reversing upon command the phase of the stnsor and second harmonic 
reference signals. This allows the detection of unexpected electronic 
offsets which could develop during op<?j"ation in orbit. 
MAGNETOMETER TRANSFER FUNCTIONS 

Since^the post-detection bandwidth of the system is much smaller 
than the passband associated with the preamplifier tunded circuits, 
the closed loop transfer function for the magnetometer (not including 
the output low-pass filter) can be well approximated by 



H(S) 



2 ^1^2 ^2"*"^^ 



(6) 



1pm 



XA G 
1pm 



where t- , is the synchronous defector time constant, t^ the integrator 
time constant, A is the preamplifier gain and K^ the sensor sensitivity 
in [Volts/Amp]; G is the equivalent transconductance of the feedback 
amplifier. In the MAGSAT design, the output filter for the synchronous 
detector is not isolated from the integrator and this causes a staall 
time constant t_ << t~ to be added to the integrator time constant in the 
second term of the denominator of (6). In analogy to a second order system. 



we can define 



„ 2 ^ J_£jn 



n 1pm 

Hence, given C and >; we ctm select the time constants t. and t„ 
to satisfy equations (7) and (8) . For this instrument the closed loop 
3 db cutoff frequency was selected as 50 Hz and X, = 0.7. The output 
filter reduces the bandvidth to 8 Hz and its response for 
frequencies near and above the cutoff is then closely approximated by 
a 6 db/oct rolloff first order low-pass filter. 

The values compv.i.ed from equations (7) and (8) ensure loop stability 
and the desired small signal frequency response. However, for large 
output amplitudes, the response may be limited by the maximum obtainable 
slew rate at the output of the integrator, 

E (synchronous detector) 

SR = -OH^JL—l (9) 

max T- 

Here E is the maximum output voltage (~ + 6 volts) which can be 
omax ' f ^ _ 

obtained from the synchronoiis detector. From equations (7) and (8) 
we can observe that large values of K.A (sensor-preamplifier gain) 
require large values of i „ (integrator time constant) for stability. 
This leads l:o reduced values of maximum slew rate, limiting the 
amplitude of hlih frequency signals that can bo followed by the instru- 
ment. On tne other hand, small values of K.A load to significant 

1 p 

and tc^nperaturo dependent zero level errors due to variations in the 



lA 



synchronous detector and integrator "effective" offset errors with 

temperature. Thus the choice of o) , C and K,A has to be carefully 

nip ■' 

analyzed for loop stability and obtainable slew rate. Large values 
of w accompanied by large values of r, (overdamped system), selected 
to obtain the desired 3 db rolloff frequency, lead to larger values 
of maximum slew rate. 

In the present design, the maximum slew rate is ~ 9000 volts/sec. 
which, if we consider the maximum output amplitude of 8 volts, corresponds 
to an output frequency of 180 Hz, well beyond the loop cutoff. 

The above considerations were taken into account in selecting the 
present implementation approach for the instrument. A magnetometer with 
a dynamic range of + 64,000 nT and equivalent zero level stability per - 
formance would have required an integrator time constant 32 times larger 
than the present design. Note that in either case the same slew rate 
in terms of field amplitude, 2.25 x 10 nT/sec, is obtained. 
FIELD OFFSET GENERATOR 



A simplified diagram of the field offset generator is shown in 
Figure 9. This is a conventional design utilizing an R-2R resistor 
ladder network. Low "ON" resistance junction field effect transistors 
with ultra-low leakage currents are utilized for the switches, while 
the operational amplifier is the sam.^ as that used in the magnetometer 
feedback loop. The ladder is used in the current mode to minimize 
common mode errors associated with the amplifier. 

The same considerations regarding thermal e.m.f.'s discussed for 
the feedback amplifier apply to this circuit. The ladder resistors are 
ultraprecision custom units matched to a tolerance of + 0.001% to simplify 
the calibration and data reconstruction procedures. 



15 



The voltage roftn-cnce was Implemented around a temporature stabilized 
precision zener (LM199A) which was subjected to a 1000 hour screening 
test for long term stability. 

The digital inputs to the offset generator swltclies are derived 
from an up-down counter associated with the 12-bit A/1) converter logic. 
When the 'fine' digitized signal for a given axis exceeds an upper or 
lower threshold (+ 1000 nT) , the counter is allowed to count up or down 
respectively, to maintain the magnetometer within its oper;'<tlng range. 
CALIBRATION AN D ALI(:N>IENT 

The problem of determining the orientation of a magnetic field vector 
has traditionally been solved by assuming that the field orientation can 
be accurately established by the geometry of a calibration coil system. 
This method is generally sufficient to determine sensor or ientat imis 
within a few minutes of arc from its true direction but It is certainly 
not adecju.ite for the present case wiiere accuracies of the order of 
2 seconds of arc are required. The metliod utilised to determine llie 
sensor alignment was similar in principle tii tliat presented by McTlierron 
and Snare exi-ept thai only two sensor rolatfons are leiiti i roil and 

tlie system of ctiuations to be solved is considerably simplified. A 
detailed description ol the modilied method has been j'.iven by Acuna 

The basic assumption made is that if the di-viations frura i^r thogonal i ty 
of the sensor assembly and triaxial test ceil ;;vstem are small (within 
a few degrees), the measurements obtained from a sensor mounleil perpendicular 
to tl\e field direction will ri-flect the sum ol the deviations ot (he sensor 
and test coil systems. 



16 



An optical reference coordinate system is first accurately established 
by means of a pair of first order theodolites which are rigidly mounted 
to suitable supports, leveled and referenced to a stable azimuth direction. 
This azimuth reference was established by means of two anchored concrete 
piers, separated by approximately 400 meters, outside the magnetics test 
facility. 

The magnetic field vectors in the reference and coil coordinate 
systems are related by a transformation matrix 

H ^ = [Bl H ., (10) 

ref coil 

and the measurements: .in the sensor coordinate system are related to the 

reference coordinate system 

M = K [A] H . (11) 

sensor ^ ref 

where M^^ ^^ represents the measured vector when a vector H ^ is applied 

in the reference system, and K is the scale factor proportionality constant. 

Thus, when the sensor is tested in the coil system, its output 

can be written as 

M = K[A] [B] H .^ (12) 

sensor coil 



and under the assumption that [A] and [B] are nearly diagonal 

M s K [ [I] + [bA] + i'[Bl ] H .^ (13) 

sensor t i j l , j ^^^-j^ 

where [OA] and [ili] are the orthogonality deviation matrices. 

In order to determine all the off-diagonal elements of [A] and [B] it is 

necessary to obtain measurements for three sensor positions differing 



17 



from the previous one by an exact 90 rotation established by the 
optical system, and by energizing each test coil system axis independently. 
The procedure thus yields not only the sensor alignment with respect to 
the reference coordinate system, but also the matrix [B] , the test coil 
system alignmefnt simultaneously. 

The advantages of this method are numerous, but the most important 
is that the test coil system is only required to be structurally stable 
for the duration of one alignment test (~ 1 hour). The ultimate accuracy 
obtainable with this method is determined by the stability and noise 
characteristics of the magnetic test facility. In the case of the 
GSFC 6.7 meter Helmholtj; coil system, the 'equivalent noise' is approxi- 
mately 1.5 seconds of arc. 

The magnetometer scale factor is calibrated using a proton precesion 
magnetometer for fields :.bove 20,000 nT. For small fields, the calibra- 
tion constants are synthesized from selected incremental measurements 
above 20,000 nT such as to determine the exact 'weights' of each of the 
7 bits in the offset generator. 
SOMMARY 

A description of the MAGSAT Precision Vector Magnetometer has been 
presented. The instrument represents the state-of-the-art in precision 
vector magnetometers and covers the range of + 64,000 nT using a 
+ 2000 nT basic magnetometer and field offset generators. The design 
utilizes ultraprecision components and electronic compensation of the 
scale factor temperature dependence by sensing the changes In resistance of tlie 
feedback coil. Brief discussions of the instrument transfer function, dynamic 
response and sensor alignment dotfrmination method were presented. A 
summary of the instrument performance characteristics is given in Table 1. 



18 



ACKNOWLEDGMENTS 

The authors wish to acknowledge the dedication and support provided 
by many of our colleagues of NSWC and GSFC during the development and 
test of the instruments. These include R„ Lundsten and John Scarzello 
of NSWC, Sanford Hinkal, Floyd Hunsaker and Carroll Fewell of GSFC and 
Northrop Services respectively, and the GSFC Magnetic Test Facility 
personnel, R. Bender, R„ Ricucci and Co HarriSo Itie support provided by 
Eo Worley and Ho Huffman in constructing prototypes and flight instruments 
is deeply appreciated. 



R EFERENCES 

[1] R. A. Langel, R. D. Reagan, J. P. Murphy, MAGSAT: A Satellite for 
Measuring near Earth Magnetic Fields, GSFC X-922-77-199, July 1977. 

[2] M. H. Acuna, Fluxgate Magnetometers for Outer Planets Exploration, 
IEEE Trans. Magnetics , Vol. MAG-10, 3, 519, September 1974. 

[3] K. W. Behannon, M. H. Acuna, L.F. Burlaga, R. P. Lepplng, N. F. Ness, 
and F. M. Neubauer, Magnetic Field Experiment for Voyagers 1 and 2, 
Space Science Reviews , 21, 235-257,. 1977. 

[4] R. L. McPherron and R. C, Snare, A Procedure for Accurate Calibration 
of the Orientation of the Three Sensors in a Vector Magnetometer, 
IEEE Trans. Geoscience Electronics , Vol. GE-16, 2, 134-137, April 1978. 

[5] M. H, Acuna, MAGSAT-Vector Magnetometer Absolute Sensor Alignment 

Determination, NASA-GSFC Technical Memorandum 79648, September 1978. 



TABLE 1 

MAGSAT - PRECISION VECTOR hUGNETOMETER 

SUMMARY OF TECHNICAL CHARACTERISTICS 

BASIC FLUXGATE MAGNETOMETER 

Dynamic Range ; + 2000 nT 

Resolution (12-blt A/P) ; +0.5 nT 

Noise ; (8 Hz BW) 0.008 nT RMS 

Zero level stability ; 

Sensor (-60°C to +60°C) +0.2 nT 
Electronics (-20°C to +50"c) + 0.2 nT 

Drive Frequency ; 12.5 KHz 

Linearity error ; < 1 x 10 

OFFSET FIELD GENERATOR: FIELD COMPENSATION L-NIT 
Dynamic Range : + 64,000 nT 
Quantization Step ; 1000 nT 
Temperature Coefficient : < 0.5 ppm/ C 
Long Term Stability ; < 2 x 10 /year 

SENSOR ASSEMBLY 

Mass : 0.6 Kg 

Dimensions : 11. A x 5.72 x 5.8 cm 

ELECTRONICS (redundant) : 

Mass : 2. 6 Kg 

Dimensions : 22.23 x 17.8 x ]1.4 cm 

Total Power Consumption : @ 25 C 1.8 watts 

@ 0°C 2.0 watts 



'/ 



ATTfTUOE tPANSFtH 
SYGTtM lAPl. JHUl 



TRIAXIAL 

FlUXGATE 

SENSOR ASSV 



I 

BOOM I L- 



FLUXGATE 
ANALOG A- 
ELECTRONICS 



X, v. Z 
»2.0O0v 



A'B REDUNDANCY SELECT 



T 



THEHMAL CONiHOL 
(APL.JHU) 



SUN SENSOR 
ASSY ;apl,JHU) 



'T 



FLUXGATE 
ANALOG B- I — I 
ELECTRONICS 



X. Y. Z 

i64.0O0y 



TIMING 6 
CONTROL 



V 



7 BIT OFFSET 
FIELD GEN. A' 



. . 



24 BIT •COARSE- 
BUFFER "A • 



- -250 KHi CLOCK 'A' 

- SAMPLE -A- 

COARSE WORD GATE 'A' 
• COARSE DATA 'A' 



12 BIT ADC 'A' 



- SHIFT CLK. A' 



40BIT FINE' 
BUFFER 'A' 



DATA PROCESSOR 'A' 



=} 



12BIT ADC 'B' 



40 BIT -FINE' 
HUfFER B- 

1 



FINE DATA 'A' 
FINE WORD GATE 'A' 
DIGITAL FLAGS 141 
FINE WORD GATE 'B' 
FINE DATA 'B' 



7BIT OFFSET 
FIELD GEN. 'B' 



l-c 



24BIT COARSE- 
BUFFER B- 



TIMING & 
CONTROL 



DATA PROCESSOR 'B' 



SHIFT CLK. -B- 

COARSE DATA 'B' 
COARSE WORD GATE 'B' 
SAMPLE -B- 
~2S0 KHz CLOCK'S- 



MAGSAT VECTOR MAGNETO(V"iTER 

BLOCK DIAGRAM 
FIGURE 1 



^E^} 



HOUSEKEEPING 

ANALOG 

SIGNAiS 



POWER 

CONVERTER 

A' 



. POWER 



^ 



CONVERTER 

■B- 



•B- PWR. IN 



MHA 6-31-77 



FINE OUTPUT 
(COUNTS) 

4095 -- 



•OUT 



[V] 



3071 -- 



2047 



I023-- 




*> B 



[nT X 1000] 






ItJi 



^-,Q,,|_J__..gZ ■ 5l4,j.9_^60 j 61 . 6 2 ^ 63 ^ 64 ^ 65 ^ 66 ^ 67 ^ 68 ^^125 126 127 
K 9_^ .1 -.| A ..r'^^., ^^. 60_^6i ^ 62_| _63 ^6_4_^ 65_ . 6_6_^6J_|_68 ^ 69 ^^,_26_^I2_7_^ 
FItLD OFFSET GENERATOR STEP (COARSE OUTPUT) 

FIGURE 2 



j^"" 



CENTER SUPPORT 




FEEDBACK 
COIL 



FIGURE 3 



c5 






MIRROR ( .750 CIBE ) 
{ TYP BOTH SIDES ) 




VECTOR MAGNEl OMETER SENSOT 

MAGSAT 



"3icf 

? -o 

. js. 
• O 

■ in 

i 



FIGURE 4 



AB 




FIGURE 5 




OFFSET 
GENERATOR 



I 



BIAS 

STEP 

. LOGIC 



CAL 



'rS OUTPUT TO 

n-BiJ A/D 
(FINE) 



250 KHz FROM S/C CLOCK 



FIGURE 6 






c > 



1.^ 



in: 



DtiHUi'SZ 



14 t] II n 10 ■ ■ 

CD40I0A 
Z 3 4 S 6 7 



CR1 CflZ 



FROW DIGITAL 
SVSTtM 
— SO KM/ 



rUP STATUS > 

tCOMMANDIIHTtnfACE) 



H4D1 : ; S1K 

; — 1 1 n 

16 M 14 1] 17 II 10 I 

CD4a49A 
t 1 3 4 s • r I 



MAGSAT-DRIVE CIRCUIT 



IE n 14 13 11 II 10 9 

CO4027A 
I 2 J 4 & S 7 I 



9 9' • 



MQ1 CKflOS H4Qt RCH05 

W02 CSR}3.MINfTAN R4Q2 RCBQ5 
«03 CSR U.-MlNiTAN n«(I3 RLR0S/RCRD5 



C404 CKR06 

C4l]b CHFIOt M'Q 

C48& CKROS 

C4a7 CHR01/X440 

C4DI CKR06 

C<)a9 CKRQ1.'X440 

C410 CKHOS 

C41I CKR05 



RAW niHos/Rcnos 

n405 RlROS'RCflOS 

R406 RIROS 

R407 RNfVbb* 

R408 mNbi 

H4aS'1 

R410 >aCH05 

R4n J 

R4t? HftUi't 
Rd13 RNN^S 
n4l4 RIHO& 




ANAIOC A/B 
SEUCTSTAfUS 
(TO t'C) 



flEVtSIOKIIBTT 



FIGURE 7(a) 



O :2i 

OS 

lO ? 
^ — 



MAGSAT VECTOR MAO^TOMETER 



CtOt 

CI or 

CK4* 
C10S 

cm 

CiQT 
C1DS 
CM 
Clio 

cm 
CV.J 
C1U' 

cm- 
ens 
CHS 
cn7 
cni 

CMS 

cm 

CJIt 
U2? 
Ci7J 

r.}2* 
cm 
ct» 
cui* 
cut 

C1I9 

cna 



SAMEASSIiaVI 




CKflOi 
CtROS XUQ 

ccnofi XMO 

ccoovrvR 

CKHO& 

CKROb 

CSF1I3'WINITAN 

CKnae 

CSfttlAllNtTAN 

cttnos 

CKRQ& 
CCR0S.>44a 

ccnofcyR 

Dfi 

CKRDf 

CXHDS 

CSRtI'MIKITAN 

CKHM 

CKhoi 

CKflOS 

f !■(,[;& 
CKHOS 

CXHOb 

CnhOi 

CCH06 xug 

CKfiOS 
CKflOi 



aior 

n>w 

Ht03 
RIM 
Rt05 
RIOG 

nio7 

R10I 
AIM 
RMO 



HIi3 
R1U 
BUS 
Hits 

Rm 

RUB 
Kits 
Rtro- 
Hill 

hi:)' 





IUIOfl«T>t&t. 
— — <*DIG SrtTHl 
— — <IDIC SVJItX) 



: IHfl 



RIRO& 

R*1NSS 
niHOS 
filflW 
RIROS 

RMKS» 




f MDl;kXUIQtHTtt;»V Rtf &fSn:w tl39v.M£ 



'tia 

RIZ] 






-Rir CNOTOOIGIUliriT 



aw 




.© 



Si 

•at' 


PAGE 



3 25 



RtOUHDtttT lAxq 



•SIltCTtaAT AQEUBLV 



FIGURE 7(b) 



GNOo- 



R 

R 

B/2 




FEEDBACK COIL 
1 R = Rod + apTJ 



FIGURE 8 




CAUB 



DIGITAL INPUTS 

FIGURE 9 















\ ■^ >?■ 






■j , 



^'•"^ ->'^^ 



- . ' -n 



iS 






irffl-^ft^1i«^n».^a^^ 



"--,> 







V- 



j!^3^ ,i: 



7 ,. ^-t^ 






»??>,;^^v"!^- 



'J «« 






jat..jatfc.g; 







T-.^ iiilinr ■^^ ^-t-U'-"t-Jg /^■^'^i 






" ~f- 



'S '-' 



•^y. 



k ■ ' 












f^ V 



1, < 



>;■ 



>?.^1 



-il "t -. 






fi'^ 






'* ?! 

- J"' ii - -,j 






4 
J